Constant power threshold—predicting maximal lactate steady state in recreational cyclists

Prolonged time trials proved capable of precisely estimating anaerobic threshold. However, time trial studies in recreational cyclists are missing. The aim of the present study was to evaluate accuracy and viability of constant power threshold, which is the highest power output constantly maintainable over time, for estimating maximal lactate steady state in recreational athletes. A total of 25 recreational athletes participated in the study of whom 22 (11 female, 11 male) conducted all constant load time trials required for determining constant power threshold 30 min and 45 min, which is the highest power output constantly maintainable over 30 min and 45 min, respectively. Maximal lactate steady state was assessed subsequently from blood samples taken every 5 min during the time trials. Constant power threshold over 45 min (175.5 ± 49.6 W) almost matched power output at maximal lactate steady state (176.4 ± 50.5 W), whereas constant power threshold over 30 min (181.4 ± 51.4 W) was marginally higher (P = 0.007, d = 0.74). Interrelations between maximal lactate steady state and constant power threshold 30 min and constant power threshold 45 min were very close (R2 = 0.99, SEE = 8.9 W, Percentage SEE (%SEE) = 5.1%, P < 0.001 and R2 = 0.99, SEE = 10.0 W, %SEE = 5.7%, P < 0.001, respectively). Determination of constant power threshold is a straining but viable and precise alternative for recreational cyclists to estimate power output at maximal lactate steady state and thus maximal sustainable oxidative metabolic rate.


Introduction
Well-being and health benefits are wellestablished goals to participate in recreational sports (Ley, 2020;Molanorouzi, Khoo, & Morris, 2015). Social recognition, improvements in self-esteem and self-confidence as well as gains in performance and pushing one's own limits are further relevant motives positively impacting compliance to exercise (Deelen, Ettema, & Kamphuis, 2018;Gill, Dowd, Williams, Beaudoin, & Martin, 1996;Kraus et al., 2017;Ley, 2020;Nikolaidis, Chalabaev, Rosemann, & Knechtle, 2019). Thus, for recreational cyclists exercising not only for health reasons but also aiming at improving individual performance, determination of appropriate training zones can be helpful for tailoring individual training programs and for achieving these goals (Jones, Burnley, Black, Poole, & Vanhatalo, 2019). The transition from sustainable to non-sustainable exercise intensities is commonly considered to differentiate between heavy and severe-intensity exercise. This corresponds to the maximal sustainable oxidative metabolic rate and therefore is widely considered a benchmark for prescribing appropriate training zones (Beneke, Hutler, Von Duvillard, Sellens, & Leithauser, 2003;Jones et al., 2019;Jones & Carter, 2000;Mann, Lamberts, & Lambert, 2013). For readability, this transition is referred to as anaerobic threshold (AnT) in the following.
All-out time trials (TT), which target at completing fixed distances in min-imum time or generating maximum work in fixed time, do not need expensive equipment and trained staff and thus can be an attractive alternative for recreational athletes. In recent years, TTs experienced a revival under the term functional threshold power, which is the average workload during a prolonged all-out test (Allan & Coggan, 2010;Borszcz, Ferreira Tramontin, & Pereira Costa, 2019;Inglis, Ianetta, Passfield, & Murias, 2020;Jeffries, Simmons, Patterson, & Waldron, 2019;Lillo-Bevia et al., 2019;McGrath, Mahony, Fleming, Raleigh, & Donne, 2021). Capability of all-out TTs for predicting performance and estimating workload at AnT was proven in numerous investigations on experienced, well-trained cyclists (Bentley, McNaughton, Thompson, & Vleck, 2001;Burnley, Doust, & Vanhatlo, 2006;Campbell, Sousa, Ferreira, Assenço, & Simes, 2007;Groslambert et al., 2004;Harnish, Swensen, & Pate, 2001;Sperlich, Haegele, Thissen, Mester, & Holmberg, 2011;Swensen, Harnish, Beitman, & Keller, 1999). Accordingly, TTs are widespread in competitive cycling. However, to the authors' knowledge, TT studies on low to moderately trained cyclists are missing. Moreover, TTs are only occasionally applied for determining AnT and for identifying training zones in recreational sports. This is probably due to following reasons. TT duration should be in the range of 30 min to 75 min since average power output (PO) conforms to output at AnT, whereas mathematical correction is required for longer shorter TTs (Borszcz et al., 2019;Ham & Knez, 2009;Harnish et al., 2001;Inglis et al., 2020;MacInnis et al., 2019;McGhee, Tanner, & Houmard, 2005;Pallares et al. 2020). Pacing prolonged TTs, however, challenges even well-trained and competitive cyclists (Borszcz et al., 2019;Foster et al., 1993;Ham & Knez, 2009; Koning de, Bobbert, & Foster, 1999;Mattern, Kenefick, Kertzer, & Quinn, 2001). This is all the more the case with recreational cyclists who usually do not have specific endurance experiences necessary to properly pace all-out TTs as substantiated by the widespread experience of practitioners that many athletes unexperienced in endurance sports have problems to pace the classical 12 min Cooper test.
Accordingly, in the present investigation TT duration was set to 30 min and to 45 min and TTs were conducted at constant PO in order to limit participants' strain and to eliminate the interfering effects of pacing ability and pacing strategy.
The aim of this study thus was to examine whether constant power threshold (CPT), operationalized as the maximum PO constantly maintainable over 30 min (CPT30) and 45 min (CPT45), is capable of accurately estimating AnT in recreational cyclists. CPT30 and CPT45 were evaluated against maximal lactate steady state (MLSS), which is commonly regarded to be one of the most accurate tools for identifying AnT (Jones et al., 2019; Legaz-Arrese, Carranza-García, Serrano-Ostáriz, González-Ravé, & Terrados, 2011; Mes-

Participants
In all, 25 recreational cyclists participated in this study of whom 22 performed all tests required, so that data of 11 female and 11 male probands were included for analysis. Participants had to meet following inclusion criteria: (1) 20-40 years old,

Measurements
Capillary blood samples (20 μl) were taken from the earlobe before the TT and from then on every 5 min and analyzed by laboratory staff using an enzymatic-amperometric method (Super GL, Dr. Müller Gerätebau, Germany). Results of the determination of blood lactate concentration (BLC) were not communicated to the participants unless they had finished their last test. PO at MLSS was defined as the highest PO at which BLC increased by no more than 1.0 mmol · l -1 between the 10th and 30th min of exercise (Beneke et al., 2003). Mean BLC (BLCmean) was the average of the 10th, 15th, 20th, 25th and 30th min. HR was continuously measured using a HR monitor (Polar Electro, Finland) and visible to the probands and noted every minute. Mean HR (HRmean) was the average from the 10th to 30th min. Cardiovascular drift (CVD) was the increase of HR from the 10th to 30th min. Rate of perceived exertion (RPE) was assessed immediately after cessation of TT on the Borg scale (6-20). In contrast to the original scale, probands were allowed to give RPE in 0.5 steps.

Statistical analysis
IBM SPSS version 27.0 was used to analyze data (IBM., Armonk, NY, USA). Normal distribution was tested using Shapiro-Wilk test. T-test and one-way repeated-measures analysis of variance (ANOVA), Bonferroni post hoc test and Greenhouse-Geisser correction were used for identifying and interpreting significant differences of the mean. Statistical significance was set at P ≤ 0.05 (*) for significant and P ≤ 0.01 (**) for highly significant. Squared eta (η 2 ; < 0.06 = small, 0.06-0.14 = medium, > 0.14 = large) and Cohen's d (d; < 0.5 = small, 0.5-0.8 = medium, > 0.8 = large) for independent and dependent variables were calculated to assess effect sizes. Coefficient of determination (R 2 ), standard error of estimate (SEE) were computed by means of single linear regression analyses. Percentage SEE (%SEE) was SEE divided by PO at MLSS, multiplied by 100. Bland-Altman analyses were applied for determining bias and 95% confidence interval (Bland & Altman, 1999). Data is presented as mean and standard deviation (SD) unless otherwise indicated.

Results
In all, 22 probands conducted two to six (median four) TTs for determining CPT30, CPT45 and MLSS. For anthropometric data and physiological parameters at MLSS see . Table 1 and 2, respectively. Repeated measures ANOVA identified significant differences between MLSS, CPT30 and CPT45 for PO (P < 0.001, η 2 = 0.295), BLCmean (P = 0.012, η 2 = 0.221), HRmean (P = 0.024, η 2 = 0.188)and RPE(P = 0.003, η 2 = 0.238) but not for CVD. Post hoc analysis revealed significant differences between MLSS and CPT30 and between CPT30 and CPT45 for PO, BLCmean, HRmean and RPE. Differences between MLSS and CPT45 were not significant (. Table 3). A workload higher than PO at CPT30 is, by definition, not sustainable for 30 min. Time to exhaustion at the TT performed 10 W above CPT30 workload was 21.0 ± 6.4 min and BLC, HR and RPE at cessation of exercise were 8.40 ± 1.36 mmol · l -1 , 176.4 min -1 and 19.6 ± 0.7. At CPT30 workload, 11 participants had to terminate exercise between the 30th and 45th min (38.6 ± 6.9 min). The other 11 probands were able to maintain CPT30 workload for 45 min, which was the predefined maximum test duration. Accordingly, CPT30 and CPT45 workload were the same and time to exhaustion thus could not be determined. At the TT performed 10 W below CPT30 workload all participants except one were able to cycle for 45min and BLCmean, HRmean and RPE were 5.19 ± 1.45 mmol · l -1 , 166.8 ± 9.4 min -1 and 16.5 ± 1.8, respectively. Differences to BLCmean, HRmean and RPE at CPT30 workload were all highly significant (P < 0.001) and effects sizes were large (d = 1.46, d = 1.08 and d = 1.18, respectively).

Discussion
Motives for participating in recreational sports are multifaceted (Ebben & Brudzynski, 2008). For recreational cyclists striving among others to push their limit and improve performance, precise determination of AnT can supply useful information for designing training programs and defining appropriate exercise intensities. Hence, the aim of the present study was to investigate whether the CPT approach can be a beneficial alternative for recreational cyclists to assess AnT. The individual decision whether a method is regarded advantageous essentially is a trade-off between accuracy, on the one hand, and effort and practicability, on the other. Thus, in the first instance accuracy of the CPT approach is compared to existing methods for predicting AnT and thereafter discussed with a focus on effort and practicability. For evaluating accuracy, predominantly studies applying MLSS were used since MLSS is regarded one of the most valid criterion standard (Jones et al., 2019;Legaz-Arrese et al., 2011;Messias et al., 2017).
Accuracy of CPT approach in comparison to literature methods R 2 for the interrelationship between prolonged TTs and MLSS range from 0.71 to 0.99 (Borszcz et al., 2019;Campbell et al., 2007;Harnish et al., 2001;Inglis et al., 2020;Lillo-Bevia et al., 2019). R 2 for the interrelation between lactate threshold and MLSS range from 0.31 to 0.90 and from 'not significant' to 0.95 for ventilatory threshold and MLSS (Figueira, Caputo, Pelarigo, & Denadai, 2008;Hauser, Adam, & Schulz, 2014;Heck, 1990;Laplaud, Guinot, Favre-Juvin, & Flore, 2006;MacIntosh, Esau, & Svedahl, 2002;Pallares, Moran-Navarro, Ortega, Fernandez-Elias, & Mora-Rodriguez, 2016;Peinado et al., 2016;Smekal et al., 2012;Van Schuylenbergh, Vanden Eynde, & Hespel, 2004;Zwingmann et al., 2019). In the present study, R 2 for the interrelationship between PO at CPT30 and PO at MLSS as well as between PO at CPT45 and PO at MLSS are R 2 = 0.99 and thus among the highest reported in the literature. Correlation parameters, however, are strongly influenced by the heterogeneity of the random sample and, furthermore, literature results are given in different units, both impeding comparison between studies. %SEE is considered to be the better parameter for comparing different methods and, thus, was predominantly applied for comparison reasons. If %SEE was

Keywords
Time trial · Anaerobic threshold · Functional threshold · Cycling · Endurance exercise not stated in literature, it was calculated from presented data.
In the present study, %SEE for estimating PO at MLSS from PO at CPT30 and PO at CPT45 are 5.1% and 5.7%, respectively. Comparison reveals that accuracy of the CTP approach is only marginally lower than that of prolonged all-out TTs performed on experienced endurance athletes, at least as good as that of lactate and ventilatory threshold concepts and critical power method and even higher than that of HR variability threshold, Conconi's HR deflection point and subjective rate of perceived exertion scores. Moreover, it is worth mentioning that in the present study four of five participants whose PO at CPT45 exceeded PO at MLSS reached a BLC plateau during the last 20 min of exercise. However, we assessed MLSS as the highest PO at which BLC increased by no more than 1 mmol · l -1 between the 10th to 30th min of exercise because it is the most common definition. If we instead defined MLSS conditions during the last 20 min of exercise, four of the five participants would have kept MLSS conditions. Bias, SEE and %SEE would have been 3.2 W, 7.3 W and 4.1%, respectively, and just in one participant CPT45 would have exceeded PO at MLSS by 10 W. Hence, the risk of prescribing too high exercise intensities and thereby to induce overloading is very low. Precision could have been further increased by applying 5 W increments but 10 W increments are standard for MLSS determination and strain for participants would have been higher. Nevertheless, data clearly point out that the CPT approach is capable of precisely determining AnT in recreational athletes and to serve as a valid cornerstone for assessing training zones. The present results are in line with a study performed on recreational runners which found %SEE to be 2.6% for predicting MLSS speed from maximum constant running veloc-

Applicability of critical power approach and all-out TTs to recreational cyclists
As mentioned before, not only accuracy but also effort and practicability are further decisive factors determining suitability of methods for determining AnT. Accuracy of HR variability threshold, Conconi test and perceived exertion approaches appears insufficient for ambitious sportsmen and -women, and lactate and ventilatory threshold concepts require expert staff and expensive equipment, which most recreational cyclists do not have access to. Therefore, discussion is focussed on applicability of all-out TTs and critical power approach. Critical power is reported to be higher than PO at MLSS so that mathematical corrections are necessary. Calculations are facilitated by customer friendly data calculators available in the internet but bear a noteworthy risk of errors since differences between critical power and MLSS range from minus 2% to plus 16.3% (Dekerle et al., 2003;Jones, 2019;Maturana et al., 2016;Okuno et al., 2011;Pallares et al., 2020). Moreover, determining critical power requires three to seven maximum-effort work bouts each inducing very high BLC and a feeling of discomfort. This is very challenging, particularly for low to moderately trained recreational cyclists who usually are not used to exercise in such high intensity zones limiting practicability for recreational athletes.
As stated before, studies investigating capability of prolonged all-out TTs for estimating AnT were almost exclusively conducted on trained cyclists. However, practicability to recreational athletes and transferability of these results is questionable due to following reasons. There is evidence that prolonged all-out TTs should last about 30 min to 75 min in order to render mathematical corrections unnecessary for calculating workload at AnT. A duration of 30 min to 75 min conforms to time to exhaustion at MLSS in recreational and trained cyclists (Baron et al., 2008;Dittrich, de Lucas, Beneke, & Guglielmo, 2014;Fontana, Boutellier, & Knopfli-Lenzin, 2009;Grossl, Dantas De Lucas, Mendez de Souza, & Antonacci, 2012) and output almost equals workload at AnT (Borszcz et al., 2019;Harnish et al., 2001;Ham & Knez, 2009;McGhee et al., 2005). In contrast, at TTs shorter than 30 min average workload exceeds workload at AnT, whereas it falls below at TTs lasting longer than about 75 min (Borszcz et al., 2019;Lillo-Bevia etal., 2019;Campbelletal., 2007;Harnish et al., 2001;Sperlich et al., 2011), so that mathematical corrections are inevitable. Allen & Coggan (2010) recommended to correct power by the factor 0.95 but this factor is challenged (Inglis et al., 2020;MacInnis et al., 2019). Hence, calculating AnT from power output of TTs shorter or longer than about 30 min or 70 min is a source of errors and derogates accuracy of AnT determination. However, outcome of prolonged TTs is strongly influenced by the athletes' ability to adequately pace TTs (Fosteretal., 1993;Ham&Knez, 2009;Koning de et al., 1999), but pacing prolonged TTs challenges even trained cyclists. Mattern et al. (2001) compared a predetermined slow starting pattern to the self-selected strategy in competitive cyclists and found the self-selected strategy to be unfavorable. Borszcz et al. (2019) investigated the relationship betweenthe functional threshold powerand PO at MLSS, but although all participants were experienced cyclists, test was preceded by a specific 50 min warm-up and must have been performed at least once before accuracy was markedly better in the well-trained than in the trained cyclists. In general, recreational athletes do not have specific experiences from TTs or competitions and thus are barely able to properly pace such trials. Finally, terrain and wind resistance strongly influence cycling velocity during field tests so that technical equipment such as power measuring devices are needed for accessing mean workload. Above that, most stationary bikes at home and in commercial fitness centers do not have the capability to compute mean workload. Due to aforementioned reasons, practicability of prolonged all-out TTs is regarded limited for recreational cyclists.

Applicability of the CPT approach to recreational cyclists
Athletes participating in the present study had only low to moderate aerobic capacities and the majority was not even practicing endurance sports. Determining workload of the first TT based on probands' self-report thus was only a first estimate and not necessarily close to MLSS. Accordingly, number of required TTs varied markedly from two to six. However, this methodology was applied in order to ensure transferability to practice of recreational sports. Procedure for determining CPT is doubtlessly very exhausting. It therefore has to be emphasized that high motivation and, in particular, good health is an inevitable prerequisite for conducting the required TTs. Moreover, a medical examination prior to testing is strongly recommended. Nevertheless, even though not encouraged during the TTs, 22 of 25 participants conducted all tests required for identifying CPT30 and CPT45. One test person had to finish testing ahead of schedule due to illness. Only two probands refused to conduct the presumably final TT for confirming CPT. They argued that they could not maintain any higher workload and CPT thus was clearly identified. Hence, the vast majority of recreational athletes participating in the study were willing and able to withstand the strain.
PO at CPT45 almost equals PO at MLSS, whereas PO at CPT30 is 3.4% higher. Accordingly, shortening TT duration is not recommendable and determining CPT45 rather than CPT30 appears appropriate for recreational athletes in order to avoid error-prone mathematical correction. Yet, some factors mitigate overall burden. First, strain decreases notably once workload falls below PO at CPT30. At a workload 10W lower than CPT30, corresponding to a reduction of about 5%, all participants were able to exercise for 45 min and RPE values decreased from 19.6 to 16.5. This conforms to the verbal feedback of most participants that at maximum two or three TTs were perceived very hard and that transition from sustainable to non-sustainable exercise occurred within a rather narrow and clearly perceptible intensity range. Second, in practice, time and effort can be further reduced. Number of TTs can be reduced by approximating AnT workload prior to CPT45 determination by means of less straining methods, such as HR variability threshold or Borg's rate of perceived exertion score. Third, CPT45 determination can be integrated into regular training programs since heavy to severe-intensity exercise bouts, which are generally an integral part of most endurance exercise programs designed for improving performance, can be replaced by TTs required for CPT45 determination. Cyclists who attach great importance to accuracy of AnT determination and are willing to accept the strain can further improve precision by determining CPT45 up to a precision of 5 W.
Furthermore, and in contrast to other methods for determining AnT, the cardiovascular drift, which is the HR increase in the course of constant load exercise (Coyle & Gonzales-Alonso, 2001), can be directly assessed during CPT determination and provide further helpful informationforcontrolling endurance exercise intensity in the field.

Conclusions
Evidence is provided that determination of CPT45 is a viable and precise approach for young healthy recreational cyclists who value accurate assessment of MLSS as point of reference for prescribing endurance exercise intensity and assessing appropriate training zones. It is almost free of costs in that it only requires a regular HR monitor and power adjustable ergometers, which are standard in most fitness centers. Some recreational cyclists even have power measuring devices at their disposal. It does not require high aerobic capacity and, in contrast to traditional TTs, no specific pacing ability. However, it has to be pointed out that determining CPT45 is very exhausting and therefore requires high motivation and foremost good health. Moreover, study was carried out on participants in the third and fourth decade of age so that further research is needed in how far this approach is applicable to other groups of athletes.